The Bitcoin equity wedge is now four research problems, not one. The accounting rule change that took effect for fiscal years beginning after December 15, 2024 made that impossible to ignore.
Only one of the four categories actually behaves like Bitcoin beta. $MSTR, $MARA, $COIN, and IBIT do not answer the same question, and treating them as interchangeable expressions of the same trade throws away most of what the filings show.
MSTR Is a Capital Allocation Machine
Start with $MSTR. The company disclosed aggregate fair market value of approximately $64.04 billion as of April 26, 2026, per the May 2026 10-Q. Trailing software revenue was $354.25 million for the period ending September 30, 2025. The operating business is a rounding error. The actual question is whether $MSTR can keep tapping capital markets at terms that let it accumulate Bitcoin faster than dilution erodes per-share exposure.
Fair-value accounting now drags Bitcoin price straight into reported earnings. A rally produces a big quarterly gain. A drawdown produces a big quarterly loss. Neither number tells you anything about the software business, and neither is the right input for modeling the equity. BTC held per share, financing cost, and the spread between implied acquisition cost and spot are what matter.
$MSTR's Filing Risk Score sits at the ceiling of the range, driven by the density of capital markets filings the company generates. That elevated disclosure cadence is itself the signal. $MSTR files constantly because it raises capital constantly. The operating-company toolkit does not work here.
Miners Hold Bitcoin but Earn Through Production
$MARA and $RIOT are a different problem. Both carry meaningful Bitcoin. $MARA disclosed aggregate fair market value of approximately $2.41 billion as of March 31, 2026, per the May 2026 10-Q. $RIOT disclosed approximately $1.07 billion as of March 31, 2026, per the April 2026 10-Q. $CLSK disclosed approximately $813.22 million as of March 31, 2026, per the May 2026 10-Q.
Those treasury positions now generate quarterly mark-to-market line items in earnings. But the positions are a byproduct of mining, not the purpose of the business. Hashrate, energy cost, and the gap between production cost per coin and the price at which coins get sold or held are the real economic engine.
$MARA reported revenue of $174.61 million for the quarter ending March 31, 2026. $RIOT reported $167.22 million for the same period. $CLSK reported $766.31 million for the period ending September 30, 2025. These are operating businesses with real cost structures. Fleet utilization, power purchase agreements, and post-halving production economics determine whether the business creates value or burns it. The Bitcoin treasury amplifies the equity's sensitivity to Bitcoin price. It does not replace the operating model.
Price action confirms the gap. $RIOT gained roughly 54% over 90 days and roughly 85% year to date through May 15, 2026, and is the only ticker in this group classified in both a short-term and long-term uptrend. $MARA is up roughly 57% over 90 days. Those moves run larger than Bitcoin itself over the same window. That is operating leverage stacked on top of Bitcoin exposure, not a tracking position.
COIN Earns Through Volume, Not Holdings
Coinbase is a third category. Its Bitcoin exposure runs through trading volume, custody fees, and the breadth of crypto activity on the platform. $COIN reported revenue of $1.41 billion for the quarter ending March 31, 2026. That number reflects how much trading happened, what mix of products customers used, and what regulators were doing to the business.
$COIN does hold Bitcoin, but the position is small relative to the revenue base. Transaction revenue as a share of total revenue, the trajectory of services revenue, and the regulatory backdrop are what move the equity. A Bitcoin rally tends to lift trading volume and therefore revenue, but the link runs through customer behavior, not a balance-sheet mark.
$COIN's 30-day price change through May 15, 2026 was essentially flat, while $MSTR gained roughly 24% and $RIOT gained roughly 35% over the same window. That divergence is the operating cycle speaking, not noise.
ETF Wrappers Are the Only Pure Beta
IBIT, FBTC, and ARKB are the one category where the Bitcoin beta framing actually fits. There is no operating business, no capital structure complexity, and no production economics. Value tracks Bitcoin price, AUM, and flows. The filing risk for all three sits in the watchlist range, reflecting routine fund reporting rather than material event density.
Performance lines up with the structure. IBIT, FBTC, and ARKB each gained roughly 5.3% over 30 days and roughly 15% over 90 days through May 15, 2026. The clustering is what you would expect from wrappers tracking the same underlying. The gap between that tight cluster and the $MSTR, $MARA, and $RIOT moves over the same period is the clearest picture of the category difference.
| Category | Examples | Dominant variable | What fair-value accounting changes |
|---|---|---|---|
| Treasury holder | $MSTR | BTC per share, financing access | Earnings now track Bitcoin price directly, burying operating results |
| Miner | $MARA, $RIOT, $CLSK | Hashrate, energy cost, production | Treasury marks add quarterly noise on top of operating results |
| Exchange | $COIN | Trading volume, services revenue | Modest holdings add minor marks, not the core model |
| ETF wrapper | IBIT, FBTC, ARKB | AUM, flows, NAV mechanics | No change, fund accounting was already mark-to-market |
The Other Reading
The strongest objection: the categories matter less than the filings suggest, because Bitcoin price dominates all of them when it counts. A 30% Bitcoin drawdown hits $MSTR on the treasury mark, hits miners on both the treasury mark and margin compression, hits $COIN on volume, and drops the ETF wrappers in lockstep. Correlation across categories spikes precisely when investors care most.
The macro tape gives that view real support. Bitcoin dominance was 58.2% at the time of this snapshot, indicating a Bitcoin-led tape where altcoin and equity correlations tend to compress. In a sharp risk-off move, the category distinctions can feel academic.
The rebuttal: correlation during stress does not collapse the categories in normal conditions, and it does not make them recover the same way or at the same speed. $RIOT's roughly 170% one-year gain through May 15, 2026 reflects operating leverage and a specific production cycle, not just Bitcoin price appreciation. $COIN's flat 30-day stretch during a period when $MSTR and $RIOT both rallied hard reflects operating dynamics that pure Bitcoin beta cannot explain. The categories converge under stress and diverge in recovery, which is exactly why conflating them before the stress event produces the wrong read on what is actually driving the equity.
Accounting Made the Problem Visible
Before fair-value accounting, the category problem was easy to ignore. Bitcoin holdings sat at cost. No quarterly earnings noise. Investors could treat every Bitcoin-linked equity as a rough proxy for Bitcoin exposure and not get caught out in any single quarter.
Now the quarterly tape forces the distinction. $MSTR's reported net income swings by billions based on where Bitcoin closes on the last day of the quarter. Miners report treasury gains and losses next to production results. The two numbers tell two different stories, and reading them as one distorts what the business actually earned.
The accounting rule did not create the category problem. It just stripped the cover off it.
Research only. Not investment advice.